Everyone prices self-hosting as the cheap tier. i'm doing the opposite — it's my most expensive option.
Quick one that's been bugging me, curious if i'm wrong.
I build infrastructure for saas (auth, billing, the operational layer). just shipped a self-hosted version — you run it on your own infra instead of my cloud.
every instinct i had said price it cheap. self-hosting feels like the "you do the work, so you pay less" option. open-source brain. the customer hosts it, runs it, maintains it — surely they pay less than the managed version, right?
the more i sat with it, the more backwards that felt.
here's the thing. self-hosting is the one thing my cloud-only competitors literally can't offer. it's the hardest part to build. and the people who actually need it — data sovereignty, regulated industries, founders who refuse vendor lock-in — those aren't the price-sensitive ones. they're the ones with the most to lose if they DON'T own their stack, and the most budget to make sure they do.
so pricing my hardest, rarest, most-defensible feature as the cheap tier is... pricing my moat like a discount. that's nuts when you say it out loud.
so i flipped it. self-hosted is now my premium option, not my budget one. cloud is the accessible entry point. self-hosting costs more, because it's worth more to the people who need it.
it also happens to be my stickiest revenue — nobody rips out infra they've embedded in their own servers — so pricing it low would cap my best customers at hobby money forever.
the part i keep second-guessing: it goes against every "self-hosting = cheaper/free" instinct in the dev world. open-source culture trained everyone to expect self-host to be the budget path.
so question:
- would you actually walk away from a tool because its self-hosted tier cost MORE than its cloud one? or does "i own it completely + no lock-in" justify paying a premium for you?
- and for anyone who's priced a self-hosted product — did you go cheap or premium, and did it work?
Wouldn't walk away — if anything, a self-hosted tier costing more signals it's actually built and supported properly, not bolted on as an afterthought. Cheap self-hosting is usually the version I'd be suspicious of.
The framing that clicks for me: you're not pricing the deployment model, you're pricing who needs it. The team that needs data sovereignty or zero lock-in isn't comparing your tier to a competitor's tier on price — they're weighing "do I own my infrastructure or not," which is a completely different budget conversation.
I went the opposite direction on what I'm building — cloud-only, no self-host option, but made the free tier genuinely usable with real limits instead of a crippled trial, and gated paid tiers on team/scale features rather than core functionality. Different lever, same underlying idea though: price what's actually differentiated, not what's cheapest for you to hand out.
We took the opposite approach with Kumiko — self-hosting is free (open source, BUSL-1.1), paid tier is the hosted platform.
Reasoning: if someone self-hosts, they already took on the ops burden. Charging extra on top felt wrong. And open-sourcing it built trust + adoption far faster than a paid self-hosted tier ever would have.
Sticky revenue comes from teams growing fast who don't want to think about infra. They're usually the ones upgrading to managed hosting without much friction.
Your moat argument still makes sense for a closed product though — "you can't get this level of self-hosted control anywhere else" is a real differentiator. Curious how conversion looks once people actually run it on their own infra.
in my case, self-hosting is like enterprise feature. which can't be free.